Create a ~/.kaggle/kaggle.json file
This is a one-time operation you need to run per kernel
Go and download your kaggle.json file from your kaggle account page and replace YOUR_USERNAME and YOUR_KEY below.
Then run this code in a kaggle python cell
text_file = open("/root/.kaggle/kaggle.json", "w")
n = text_file.write('{"username":"YOUR_USERNAME","key":"YOUR_KEY"}')
text_file.close()
!chmod 600 /root/.kaggle/kaggle.json
# Make sure it worked
!cat /root/.kaggle/kaggle.json
Enable internet access
Go to your kernel settings and enable internet access.
You might need to verify your account before you do it (get an SMS typically)
Create the predictions
Example prediction generation:
from sklearn.ensemble import RandomForestClassifier
import datetime
features = ['Pclass', 'Sex', 'Age', 'SibSp', 'Parch', 'Fare', 'Embarked']
y = train_set["Survived"]
X = train_set.drop(columns=['Survived'])
X_test = test_set
model = RandomForestClassifier(n_estimators=100, max_depth=5, random_state=1)
model.fit(X, y)
predictions = model.predict(X_test)
predictions = predictions.astype(int)
output = pd.DataFrame({'PassengerId': p_test.PassengerId, 'Survived': predictions})
file_name = f'submission_{datetime.datetime.now().strftime("%Y-%m-%d-%H-%M-%S")}.csv'
output.to_csv(file_name, index=False)
print("Your submission was successfully saved: " + file_name)
Submit
And then in another cell (or in the same cell, that's ok) run this:
submission_message = f'Features: {", ".join(features)}'
# Submit
!kaggle competitions submit -c titanic -f {file_name} -m "{submission_message}"
# List submissions
!kaggle competitions submissions -c titanic